Evaluative Movement Assessment - EMA

Technical Manual

Script Author: Katja Borchert, Ph.D. (katjab@millisecond.com), Millisecond

Created: January 03, 2016

Last Modified: January 03, 2025 by K. Borchert (katjab@millisecond.com), Millisecond

Script Copyright © Millisecond Software, LLC

Background

This script implements a version of the Evaluative Movement Assessment (EMA) procedure. The EMA is a response time measure of implicit evaluations. Specifically, this script is designed as a template and assesses people's implicit attitudes towards "weekend", "mosquito", and "airport" using well established EMA targets "cavities" and "movie theaters" as attitude calibration targets.

It is suggested that the calculated EMA scores for several target categories 1) help rank the evaluations on a single scale 2) provide a neutral preference point for the scale

References

Miguel Brendl, C.M, Arthur B. Markman, A.B. & Messner, C. (2005). Journal of Experimental Social Psychology, 41, 346–368.

Duration

30 minutes

Description

Participants are asked to move words (targets and distractors) -that are displayed either to the right or left of their own name- either towards or away from the name. In this script, participants move the stimuli by repeatedly pressing a left or right response button which 'moves' the stimuli visually closer or father away from the name. Targets: all towards (or all away; depending on block instructions) Distractors: towards/away depends on valence of word

Procedure

1 practice block of 24 trials: only distractor items (12 positive, 12 negative; randomly selected from the list of distractors)
5 EMA blocks run in a AABBA/BBAAB within subjects design (performance during the first A block is not considered for calculation of EMA scores)
(order is counterbalanced by groupnumber)

Each EMA block: 83 trials
- 5 Targets tested (each with one target word) - each repeated 7 times = 35 target trials
the odd number of trials results in the targets being unequally often left or right to the name.
For each participant, it's decided randomly whether the targets are more often left than right to the name.

- 40 distractor items (20 positive, 20 negative) - each repeated once = 40 distractor trials
- 4 filler items (2 positive, 2 negative) - each repeated 2 times (used as prefatory trials before start of block and after rest period) = 8 filler trials
- each block presents a rest trial in the middle (block starts again with 4 filler trials)

Block Sequence:
- 4 filler trials
- 38 randomly selected trials from targets, distractors
- rest trial
- 4 filler trials
- 37 randomly selected trials from targets, distractors

Trial Sequence:
blank (600ms)->name in white frame + 4 XXXX (700ms)->XXXXmask replaced by red attitude word-> wait for response -> potential error feedback


EMA Score Calculation: EMA scores are based on the difference in median latencies of away and toward movements corrected by a calibration constant

Positive Response Bias (positivity bias): participants tend to be faster in moving stimuli towards their names than away from their names
regardless of valence of target words
=> Diff (awayRT - towardsRT) tend to be positive (which affects the neutral point evaluation though not the relative ranking of 'preferences')
=> correction/calibration necessary

Calibration Algorithm applied in this script:
1) run known (2) target words as calibration targets; these target words need to roughly have a mean valence rating of 0
(here: 'movie theater' and 'cavities', see Brendl et al, 2005)

2) establish calibration constant: CC = meanRT(away) - meanRT(towards) for the two calibration targets
(if no bias towards 'towards' movements, one would expect CC = 0 with the selected calibration targets)

3) Use calibration constant in EMA score calculations:
EMA(target1) = median(away_target1) - median(towards_target1) - CC
EMA(target2) = median(away_target2) - median(towards_target2) - CC
EMA(target3) = median(away_target3) - median(towards_target3) - CC

EMA(target4) = median(away_target4) - median(towards_target1) - CC (=calibration target) => 0
EMA(target5) = median(away_target5) - median(towards_target1) - CC (=calibration target) => 0

based on: (awayRT-CC/2) - (towardsRT + CC/2) = awayRT - towardsRT - 2*CC/2 = awayRT - towardsRT - CC

Removal of error trials for calculation of EMA scores:
- only latencies of correct initial responses within the set response window (between 100-3000ms after onset of words)
and without response reversals are considered for EMA score calculations

Stimuli

see section Editable Stimuli

Instructions

provided by Millisecond as htm-files (htm files can be edited with Notepad (Windows) or TextEdit (Mac))

Summary Data

File Name: evaluativemovementassessment_summary*.iqdat

Data Fields

NameDescription
inquisit.version Inquisit version number
computer.platform Device platform: win | mac |ios | android
startDate Date the session was run
startTime Time the session was run
subjectId Participant ID
groupId Group number
sessionId Session number
elapsedTime Session duration in ms
completed 0 = Test was not completed
1 = Test was completed
propCorrectAll Overall proportion correct for the 4 main blocks based on performance in distractor, targets and calibration target trials
medianRTAll Overall median latency (in ms) of correct responses for the 4 main blocks based on distractors, targets and calibration trials
meetsExclusion Returns 1 if participant's error rate exceeds parameters.exclusionCriterium (default: errorate > 0.07)
cc Calibration coefficient as the difference of the two to estimate the positivity bias of the participant
positivity bias: the tendency to be faster when moving a word towards the name than away from it
(even if both directions are congruent with the valence of the word)
calculated as the difference between the mean of the away latencies of the calibration targets
and the mean of the toward latencies of the calibration targets (here: target 4 and target 5)
medianRTTarget1Away Median latency (in ms) of correct responses for moving target1 away from the name
medianRTTarget1Towards Median latency (in ms) of correct responses for moving target1 towards the name
emaTarget1 EMA score for target1; difference in away and towards median latency corrected by the calibration coefficient
medianRTTarget2Away Median latency (in ms) of correct responses for moving target2 away from the name
medianRTTarget2Towards Median latency (in ms) of correct responses for moving target2 towards the name
emaTarget2 EMA score for target2; difference in away and towards median latency corrected by the calibration coefficient
medianRTTarget3Away Median latency (in ms) of correct responses for moving target3 away from the name
medianRTTarget3Towards Median latency (in ms) of correct responses for moving target3 towards the name
emaTarget3 EMA score for target3; difference in away and towards median latency corrected by the calibration coefficient
medianRTTarget4Away Median latency (in ms) of correct responses for moving target4 away from the name
medianRTTarget4Towards Median latency (in ms) of correct responses for moving target4 towards the name
emaTarget4 EMA score for target4; difference in away and towards median latency corrected by the calibration coefficient
medianRTTarget5Away Median latency (in ms) of correct responses for moving target5 away from the name
medianRTTarget5Towards Median latency (in ms) of correct responses for moving target5 towards the name
emaTarget5 EMA score for target5; difference in away and towards median latency corrected by the calibration coefficient

Raw Data

File Name: evaluativemovementassessment_raw*.iqdat

Data Fields

NameDescription
build Inquisit version number
computer.platform Device platform: win | mac |ios | android
date Date the session was run
time Time the session was run
subject Participant ID
group Group number
session Session number
blockCode Name of the current block
blockNum Number of the current block
trialCode Name of the current trial
trialNum Number of the current trial
countBlockTrial TrialCounter per block
stimType "target1"/"target2"/"target3"/"target4"/"target5", "distractor", "calibration","filler"
direction Stores the direction in which the current word should be moved in relationship to the name ("towards" vs. "away")
stimPosition "left": word is presented on left side of the screen; "right": word is presented on right side of the screen
stimX Stores the x coordinate of the stimulus
correctMovement Stores the direction the current word should be moved ("left" vs. "right")
valence Stores the valence of the currently presented word ("positive", "negative", "")
stim Stores the currently presented word
acc Stores whether the initial movement is in the correct direction (irrespective of latencies and later movement changes)
rt Stores the latency (in ms) of the initial movement measured from onset of words to pressing the response button
valid 1 = initial response was correct, within the responsewindow, and movement direction stayed the same throughout
0 = otherwise
initialMove Stores the direction of the initial movement (left vs. right)
directionChange True = participant changed movement direction; false = participant did not change movement direction
testData 0 = data not considered as test data (= practice); 1 = data considered as test data

Parameters

The procedure can be adjusted by setting the following parameters.

NameDescriptionDefault
responseKeyLeft The left response button "A"
responseKeyRight The right response button "L"
nameSize The fontheight in canvas height percentages of the name stimulus (default; 4%)
nameLeftx The left horizontal screen position in canvas width percentage of the name/stim 30%
nameRightx The right horizontal screen position in canvas width percentage of the name/stim 70%
blankDuration The duration of the blank screen (in ms) at beginning of trial 600ms
xxxxDuration The duration of the XXXX mask (in ms) 700ms
mInvalidResponseLatency The duration of the minimum valid response latency in ms; measured from onset of targets 100ms
maxValidResponseLatency The duration of the maximum valid response latency in ms; measured from onset of targets 3000ms
iti Intertrial interval in ms 0
exclusionCriterium Sets the overall error rate in the testblocks that would exclude a participant from further analyses
if the error rate indicates exclusion from analyses,
the summary data file stores a '1' under expressions.meetsExclusion
0.07